Re: -fexpensive-optimizations + haifa

[Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>]

anthony,

I already reported this bug a few days ago, egcs even contains a testcase 
(gcc.c-torture/980511-1.c) now to trigger this bug in the combiner. Jim Wilson
is working on a solution.

Fortunately it looks like this one and the optimizer bug in fs/namei.c (mknod,
egcs seems to loose tracking info for clobbered registers across a lot of
function calls) are the last ones triggered by a kernel compile. 

Bye, Franz.

Am Sat, 23 May 1998 schrieb anthony tong:
>I've found an error compiling the vger linux kernels (2.1.103), with
>the following:
>/usr/lib/gcc-lib/powerpc-unknown-linux-gnulibc1/egcs-2.91.29/specs gcc
>version egcs-2.91.29 19980517 (gcc2 ss-980502 experimental)
>
>This is on a linux powerpc platform, built with haifa. Using -O2
>-fno-expensive-optimizations fixes the internal compiler error. It's
>a stage2 compile.
>
>gcc -D__KERNEL__ -I/usr/src/cvs/linux/include -Wall -Wstrict-prototypes
>-O2 -mcpu=604e -D__powerpc__ -fsigned-char -msoft-float -pipe
>-fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring   -c -o
>super.o super.c
>
>../../egcs/gcc/haifa-sched.c:7577: Internal compiler error in function
>schedule_region
>
>Good luck,